Page 6 of 18 FirstFirst ... 4567816 ... LastLast
Results 51 to 60 of 180

Thread: HOWTO: Make festival TTS use better voices (MBROLA / CMU / HTS)

  1. #51
    Join Date
    Aug 2008
    Beans
    1

    Re: HOWTO: Make festival TTS use better voices (MBROLA / CMU / HTS)

    Thanks for this guide MonkeeSage, I'm not sure I would have been able to get festival up and running so quickly without it.

    I have a question about the output quality of the nitech HTS voices, I got and installed the latest 2.1 voices just today but I noticed that they have a very "buzzy" quality when they're output and when I hear them on Edinburgh server they don't have this problem (http://www.cstr.ed.ac.uk/projects/fe...nlinedemo.html). Its almost as if the "STRAIGHT vocoder" described in the "An Overview of Nitech HMM-based Speech Synthesis System for Blizzard Challenge 2005" paper is not functioning. Am I missing something or perhaps I should try to build the entire system from scratch using the sources?

    Edit: Here are some audio examples so you can hear the problem that I'm talking about.
    Using the Nitech SLT voice from the online demo I get:
    http://www.arondennen.com/files/onlineSLT.wav
    which is how I would like my system to sound.

    However when using the 2.1 version of the Nitech SLT voice on my system I get this buzzy output:
    http://www.arondennen.com/files/problematicSLT.wav

    Thanks
    Last edited by tabarin; September 1st, 2008 at 11:26 PM.

  2. #52
    Join Date
    Sep 2008
    Beans
    1

    Re: HOWTO: Make festival TTS use better voices (MBROLA / CMU / HTS)

    Thank you for the usefull post.

    But does any one knows if there is posibility to make festival read romaji

    Thanks in Advance

  3. #53
    Join Date
    Apr 2006
    Beans
    66

    Re: HOWTO: Make festival TTS use better voices (MBROLA / CMU / HTS)

    I have gotten everything to work according to the howto but attempting to change the default voice using
    Code:
    (set! voice_default 'cmu_us_bdl_arctic_clunits)
    in /usr/share/festival/festival.scm gives me the following error.
    SIOD ERROR: unbound variable : cmu_us_bdl_arctic_clunits
    closing a file left open: /usr/share/festival/init.scm
    festival: fatal error exiting.
    Can someone help me with this?

  4. #54
    Join Date
    Aug 2008
    Beans
    4
    Distro
    Ubuntu 8.04 Hardy Heron

    Re: HOWTO: Make festival TTS use better voices (MBROLA / CMU / HTS)

    Quote Originally Posted by bobonthenet View Post
    I have gotten everything to work according to the howto but attempting to change the default voice using
    Code:
    (set! voice_default 'cmu_us_bdl_arctic_clunits)
    in /usr/share/festival/festival.scm gives me the following error.


    Can someone help me with this?
    Well, mine works if I put the line in voice.scm and/or siteinit.scm

    And mine is called voice_cmu_us_rms_arctic_clunits, but I might not have followed the directions exactly...I don't really remember.

  5. #55
    Join Date
    Apr 2006
    Beans
    66

    Re: HOWTO: Make festival TTS use better voices (MBROLA / CMU / HTS)

    Quote Originally Posted by BillaMonster View Post
    Well, mine works if I put the line in voice.scm and/or siteinit.scm

    And mine is called voice_cmu_us_rms_arctic_clunits, but I might not have followed the directions exactly...I don't really remember.
    That didn't work but you did steer me in the right direction and I got it working. There is a line in init.scm that does it. Its set to voice_default so I changed it to voice_cmu_us_bdl_arctic_clunits and it worked like a charm. Thanks a ton I don't think I could have figured it out on my own, since I would have had to look at all of the files not knowing what any of them do.

  6. #56
    Join Date
    Jul 2008
    Beans
    1

    Re: HOWTO: Make festival TTS use better voices (MBROLA / CMU / HTS)

    great thread... i was able to get the voices working and its cool to hear the new voices through icecast (using liquidsoap)

    now i should check out your "ghetto" front-end... maybe we could eventually get a few ppl together to set up a more complete front-end.

    thanks for the howto

  7. #57
    Join Date
    Sep 2008
    Beans
    2

    Smile Re: HOWTO: Make festival TTS use better voices (MBROLA / CMU / HTS)

    Thanks a bunch for this post. I have been playing around with festival for a while. I wasn't all that happy with the default ked voice. These new voices are much better. Everything worked great on my system. Ubuntu Hardy, festival 1.96:beta.

    I also have a quick question. I have been playing around with python using festival. Right now my programs are issuing console commands to get festival to speak: echo "some text" | festival --tts. Does anyone know how to issue text directly to festival from python rather than issuing console commands?

  8. #58
    Join Date
    Sep 2008
    Beans
    2

    Re: HOWTO: Make festival TTS use better voices (MBROLA / CMU / HTS)

    Oh one more thing. When festival is done speaking it iterates a tiny bit of the first word. Any idea what's going on and how I can fix it?

  9. #59
    Join Date
    Oct 2006
    Beans
    414

    Re: HOWTO: Make festival TTS use better voices (MBROLA / CMU / HTS)

    I've installed the voices correctly (Mbrola and HTS) yet whenever I select voice I can only get the default voice.
    Code:
    festival> (set! voice_default 'voice_us3_mbrola)
    voice_us3_mbrola
    festival> (SayText "Hello")
    #<Utterance 0xb696d268>
    festival> (set! voice_default 'voice_cmu_us_kal_com_hts)
    voice_cmu_us_kal_com_hts
    festival> (SayText "Hello")
    #<Utterance 0xb69a3d68>
    festival> (set! voice_default 'voice_nitech_us_awb_arctic_hts)
    voice_nitech_us_awb_arctic_hts
    festival> (SayText "Hello")
    #<Utterance 0xb69bb0d8>
    festival> (voice.list)
    (us2_mbrola
     us3_mbrola
     us1_mbrola
     rab_diphone
     cmu_us_kal_com_hts
     nitech_us_rms_arctic_hts
     cstr_us_ked_timit_hts
     nitech_us_bdl_arctic_hts
     nitech_us_awb_arctic_hts
     nitech_us_clb_arctic_hts
     nitech_us_jmk_arctic_hts
     nitech_us_slt_arctic_hts)
    Am I doing something wrong? Where are the .scm files?

  10. #60
    Join Date
    Mar 2007
    Beans
    7

    Re: HOWTO: Make festival TTS use better voices (MBROLA / CMU / HTS)

    Wow this works great!

    The best is the Nitech 'bdl' voice. It's nice and clear 'tenor' voice.

    Thanks for the festival.scm parameters. I wrote a simple bash script that is executed as a cron task and it tells me the time every hour. Good for my time based work that I do.

    One thing though, how do you increase the amplitude of the voice when using alsa to output the voice? I have music on sometimes and it's hard to hear the voice over the music.

    Thanks billions of times!

Page 6 of 18 FirstFirst ... 4567816 ... LastLast

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •